Fluke PM3384A combiscope trace fault

[Guineafowl]

TL;DR Summary

This scope is switchable between analogue and digital. The analogue is fine on all four channels, but the digital versions are off-screen.

I’ve just bought this useful looking 100 MHz scope in the hope of repairing it.

When I switch to digital mode, the trace disappears off-screen on all four channels. 1 and 3 are up high, 2 and 4 down low. Sometimes the top or bottom of the trace is visible, and it flickers up and down the screen. The CAL square wave, which displays fine in analogue mode, also becomes distorted, with lots of roll-off.

The autocal feature seems to freeze after a while, and not complete, but the other self tests pass.

Any ideas? I guess the input board is OK, and there is a fault with the digital board. I’m in the process of checking the power supply caps.

 

[berkeman]

Do you have the service manual for it? I tried a Google search, but got mostly fluff I think:

https://www.bing.com/search?q=Fluke...-28&sk=&cvid=D5A093A41F5B44309E34BD9CCADCB662
If you can do a firmware upgrade, that may fix it. Does it have an external port for remote operation that can be used for a firmware upgrade?

(That reminds me, I have a LeCroy 'scope in the lab that I need to fix with a firmware upgrade -- thanks for the reminder...)

 

[Tom.G]

Here is a link for a free download of the 400+ page PM3384 Service Manual. It includes schematics, parts list, board layouts, etc.
https://www.opweb.de/download.php?filename=Fluke+and+Philips--PM3394--service--ID4778.pdf
This was the 5th in the search engine list that @berkeman posted above.

Cheers,
Tom

 

[Guineafowl]

Thanks both. It does have an rs 232 port, but I’ve had no luck finding firmware.

I also have the service manual (from elektrotanya.com, a very useful source of obsolete manuals).

Once the power supply is checked, I was going to try to trace the input signal as it leaves the acquisition board and enters the digital board, to see where the distortion starts happening.

On page 417 of Tom’s manual, section, 8.11.4.7, there is a description of how to read and adjust the DAC outputs, some of which seem useful (offset CH1 and suchlike) but I can’t make head nor tail of it yet.

A factory reset would be nice, but doesn’t seem to be mentioned. Just in case a ‘phantom’ has been fiddling about.

 

[Guineafowl]

I’ve looked a little further, but lack experience in this higher-end stuff.

The square wave CAL signal makes it onto the digital board, via two coax cables (highlighted):

... and through the MASPU-A (main acquisition sig processing unit) - output looks like this:

... and through a nasty-looking hybrid buffer chip, output, called ADCOUT:

Then onto ADC-A.

 

[Guineafowl]

ADC-A is numbered OQ0239.

Now here’s where I’m stuck, because the output will be data from pins 1-10. What should data look like? I get this on all pins:

... which doesn’t look very useful, nor does it change with an altering signal in the front end, or even turning the channel off.

The ADC is heat-sinked on top, and gets too hot to touch. Probe is on x1 in all photos.

 

[Averagesupernova]

I would work with a varying DC signal on your input. It would be easier for me to wrap my head around after digitization. Just something you can vary manually like a potentiometer. I used to work on scopes for a living as well as a bunch of other test equipment. But never anything that digitized except for the DVM portion.

 

[Tom.G]

... which doesn’t look very useful, nor does it change with an altering signal in the front end, or even turning the channel off.

The ADC is heat-sinked on top, and gets too hot to touch.

That scope signal looks like an open circuit, mainly noise pickup. And the "too hot to touch" (above 125F) indicates a very sick IC. How is the temperature of the following chip, the ECL_TTL TRANSLATOR?

Now here’s where I’m stuck, because the output will be data from pins 1-10. What should data look like?

Those pins carry the Binary digital representation of the input signal level. D0 is the least significant bit, D7 the most significant bit. From the control signals and overall circuit that looks like a Flash Converter, the output should update on every CLK pulse. Verify that there is a CLK signal on pin 14 of the ADC. (It's likely to be fast and at ECL logic levels. See: https://uk.farnell.com/emitter-coupled-logic-definition)

Note that ADC-A is implemented in ECL logic for speed, and uses Gnd and a -5.2V power supply. How do the temperature and signals compare with the other ADC, ADC-B?

@Averagesupernova has a good suggestion of using a controllable DC voltage on the input... might be easier to tell if anything is getting thru that ADC.

Cheers,
Tom

 

[Guineafowl]

How is the temperature of the following chip, the ECL_TTL TRANSLATOR?

Slightly less hot. All the chips seem to run rather hot on this thing.

Those pins carry the Binary digital representation of the input signal level. D0 is the least significant bit, D7 the most significant bit. From the control signals and overall circuit that looks like a Flash Converter, the output should update on every CLK pulse. Verify that there is a CLK signal on pin 14 of the ADC. (It's likely to be fast and at ECL logic levels. See: https://uk.farnell.com/emitter-coupled-logic-definition)

^^ this is all correct, according to the manual.

It’s hard to get to any of this stuff because the boards are upright and close together. I managed to solder a jumper onto pin 25 of the ECL/TTL translator, which gets the same signal. No clock signal obvious. Following signal MCAA back, I find that it originates at the ‘whistler chip’:

... and I can’t find a clock signal at the actual crystal pins, either. Now, my scope is only 20MHz and the crystal is 200MHz. But I’d expect to see some representation of the too-fast clock signal. A Gould 4074 100Mhz scope is on the way, so watch this space. Is it likely the crystal is bad?

Note that ADC-A is implemented in ECL logic for speed, and uses Gnd and a -5.2V power supply. How do the temperature and signals compare with the other ADC, ADC-B?

ADC-B is much the same.

 

[Guineafowl]

To add to that: The crystal is labelled “G8025, 200M Hz” on the circuit diagram, but the device itself has 200000.0, TQX3828, 03TQG96.

Is it really 200Mhz? Seems hard to find a replacement.

 

[Averagesupernova]

Do you have access to a receiver that can verify a signal around 200 Mhz? A scanner perhaps? I would just tune a spectrum analyzer around 200 and look for it but I'm guessing you don't have access to one. Does the clock feed anything else that you could verify is working? Or, alternatively you can likely kill the oscillator touching the right pins with your fingers and verify if something else quits working. I would not count on seeing a 200 Mhz signal with a 20 Mhz scope.

 

[Guineafowl]

Do you have access to a receiver that can verify a signal around 200 Mhz? A scanner perhaps? I would just tune a spectrum analyzer around 200 and look for it but I'm guessing you don't have access to one. Does the clock feed anything else that you could verify is working? Or, alternatively you can likely kill the oscillator touching the right pins with your fingers and verify if something else quits working. I would not count on seeing a 200 Mhz signal with a 20 Mhz scope.

The Gould 4074 100Mhz might be able to see something, but it’s not here yet.

I did think of using the Fluke scope to probe its own oscillator signal as it works fine in analogue mode. Not sure if the oscillator would be active in this state, though. But the digital board is still used to draw on-screen text.

I have an old signal generator for aligning vintage radios that will put out 100Mhz. Perhaps I’ll use that to see what such a signal looks like on a 20Mhz scope.

 

[Averagesupernova]

If you are not sure the oscillator is active in analog mode, it brings up the thought that maybe it is not active in digital mode because it's not enabled. If the controls of the scope shut it down at times, maybe there is a control issue. Why not try to drive that xtal with your 100 Mhz signal generator and see what happens? Couple the generator output through an appropriate sized capacitor right onto one of the pins that connect to the xtal. So what if you drive it at half the frequency it normally runs at or less. If you observe a significant change in operation you'll know you are onto something.

 

[Guineafowl]

I would not count on seeing a 200 Mhz signal with a 20 Mhz scope.

Spot on. I used the Fluke itself to show a 200MHz signal from the generator, then probed the crystal on the same settings and saw a clock signal.

Looking at signal voltage levels:

... the signal coming in from the left should be -8.5V, and it is. After the MASPU, it should be:
BOOT 68: -0.5V
BOUT 65: <300mV

But BOOT 68 is is at 2.46V.

Also the ANINA signal from BUFFER pin 3 is at -5.67V not -0.5V. By biasing the level with a suitable resistor to ground, I can get a trace on the screen.

So it looks like the MASPU is the start of the signal level problem.

On this chip, pin 19 (VEETWE) should be -12V, but was -5.8V. I found this to be caused by a 1R resistor that had gone high to 16R. This didn’t solve the signal problem, though. There may be another voltage level wrong but I can only get to certain pins due to the orientation of the board. Pins checked have been highlighted.

 

[Tom.G]

Just a few obvious items to check:

• The AC/DC voltages levels shown on the schematic for BOOT & BOUT are probably METER readings; I would expect pulses on at least one of them.
• Matching voltage levels and waveforms in the "B" channel.
• Check the supplies with a 'scope to make sure they are clean.
• Disconnect the BOOT & BOUT pins to get an idea where the bad voltage originates. (you will probably need an appropriate pull-down/pull-up resistor to get an informative reading. Value depends on what logic family those chips are. Looks like they are mixed logic with level-shifting built in.)

Cheers,
Tom

 

[dlgoff]

It’s hard to get to any of this stuff because the boards are upright and close together.

I guessing that these circuit boards connect to a mother board via an edge connector. If so, then you should get an edge connector extension board like this one from https://www.digikey.com/product-detail/en/vector-electronics/3690/V1024-ND/38898&?gclid=EAIaIQobChMIy7Ko4pui4wIVSrzACh1TTwMJEAQYAyABEgLQ0vD_BwE.